Skip To Content

Set up TensorFlow deep learning framework for ArcGIS

You can work with a number of deep learning frameworks in ArcGIS, including TensorFlow.

Install TensorFlow in ArcGIS Pro

TensorFlow can be installed through a Python environment in ArcGIS Pro.

Run TensorFlow on the CPU

You can install TensorFlow to run on your computer's CPU. For more information on how to create and clone Python environments and install Python packages, see Python Package Manager.

  1. To create an environment, in the Python backstage, click the Manage Environments button, click an environment, and click the environment's Clone button.
  2. Once the environment has been created, to activate it, click the option next to the environment, and click OK.
  3. Click the Add Packages button, search for tensorflow-mkl, and install the most recent version of the package.

Run TensorFlow on the GPU

You can install TensorFlow to run on your computer's GPU. To run TensorFlow on your GPU, you need to install NVIDIA GPU drivers, CUDA Toolkit, and CuDNN SDK on your machine. See the TensorFlow software requirements for installation details. For more information on how to create and clone Python environments and install Python packages, see Python Package Manager.

  1. To create an environment, in the Python backstage, click the Manage Environments button, click an environment, and click the environment's Clone button.
  2. Once the environment has been created, to activate it, click the option next to the environment, and click OK.
  3. Click the Add Packages button, search for tensorflow-gpu, and install the most recent version of the package.

Verwandte Themen